Let's talk

Agentic AI Architect

AI Startup in Palghar, Maharashtra

Operating from Palghar, Divyaprakash Jha leads two AI companies—GodEngine and AskShiva—using a fleet of AI agents to build decision intelligence and business automation tools.

By Divyaprakash JhaUpdated 4 min readTopic: Founder

Photo for AI Startup in Palghar, Maharashtra

Divyaprakash Jha is the founder of two AI companies, GodEngine and AskShiva, operating from Palghar, Maharashtra. The studio specializes in agentic AI architecture, utilizing a fleet of fifteen AI agents directed by three human operators to build decision intelligence and business automation tools.

Leading AI Innovation from Palghar

Operating from the town of Palghar in Maharashtra, the studio represents a unique model of high-tech development outside of traditional metropolitan hubs. The founder, Divyaprakash Jha, has established two distinct private limited companies: GodEngine and AskShiva.

The studio's operational model is highly efficient. It consists of a core team of three human operators who direct a specialized fleet of fifteen AI agents. This setup allows the team to ship complex software products that would typically require a much larger traditional agency.

  • Location: Palghar, Maharashtra, India.
  • Founding Year: 2026 (full independence).
  • Core Focus: Agentic AI platforms and decision intelligence engines.

Two Companies, Two AI Platforms

The founder's work is split across two primary platforms, each solving a different problem in the AI landscape. While one focuses on high-level reasoning for complex choices, the other focuses on the practical, day-to-day needs of small businesses.

GodEngine: Decision Intelligence

Most AI tools are designed to answer questions, but GodEngine is built to help users make decisions. It is a decision intelligence engine that uses reasoning, causal links, and scenario modeling to provide structured reports. Instead of a simple chat reply, it provides a model of the world, showing the facts, the causes, and the potential outcomes of a choice.

AskShiva: Business Automation

AskShiva is an AI assistant designed specifically for the realities of Indian business workflows. It understands that an Indian business owner doesn't just need a paragraph of text; they need a finished task completed in the right format and language.

AskShiva excels at tasks such as generating GST invoices, translating messages into Marathi or Hindi, and managing follow-ups on WhatsApp. It bridges the gap between "answering a question" and "finishing the work."

Feature GodEngine AskShiva
Primary Goal Decision Intelligence Task Completion & Automation
Core Output Causal reports and scenarios Invoices, documents, and messages
Ideal User Founders, analysts, and researchers SME owners and Indian businesses

For those interested in the underlying infrastructure of these agents, Shiva Agents provides the platform for running these agents as an always-on company.

The Agentic Architecture Approach

The studio does not just use AI; it builds with an "agentic" mindset. This means the AI is designed to act as a collaborator that can plan, use tools, and verify its own results. This is achieved through a rigorous "plan, act, verify" loop.

To reflect the specialized nature of these digital workers, the agents are named using Sanskrit or Hindi terms that describe their specific craft. This gives each agent a distinct persona and role within the workflow.

The Agentic Fleet

The studio uses a variety of specialized agents to build software. Examples include:

  • Vastu: Focused on architecture.
  • Sutra: Focused on writing code.
  • Ganit: Focused on numbers and mathematics.
  • Vani: Focused on words and language.
  • Parakh: Focused on testing and verification.

This architecture allows agents to delegate tasks to one another, consult specialists, and work in parallel to complete complex, multi-step workflows.

Independent Infrastructure and Hardware

A key competitive advantage for the studio is its approach to hardware. Rather than relying solely on expensive, third-party APIs for every task, the studio runs its own GPU and CPU servers using open models.

This independent infrastructure is critical for several reasons:

  1. Cost Efficiency: By running bulk AI work on local hardware, the studio avoids the high costs associated with massive API usage.
  2. Scalability: It allows for more predictable and efficient scaling of heavy AI tasks.
  3. Reliability: It reduces dependence on single-provider uptime, as the studio can route tasks to different models or local hardware.

BusinessOS: The AI-Native Operating System

Most small and medium enterprises (SMEs) suffer from "fragmented tool syndrome." One business might use a spreadsheet for leads, WhatsApp for customers, and email for documents. These tools do not talk to each other, creating data silos and security risks.

BusinessOS is designed to be the foundation that unifies these pieces. It is an AI-native operating system that provides a single identity and permission engine for an entire company. Instead of bolting products together, it builds a master architecture first, then adds modules on top.

Key Features of BusinessOS:

  • Unified Identity: One sign-in and one set of permissions across all business functions.
  • Secure Data Isolation: Uses database-level isolation to ensure different organizations or departments cannot access unauthorized data.
  • Auditability: An append-only, searchable audit log that tracks every change and action.
  • Integrated Modules: Starts with a CRM and expands to include files, search, and notifications in one place.

The goal of BusinessOS is to move businesses away from disconnected tools and toward a single, secure, and intelligent digital workspace.

From Trading Systems to AI Agents

The transition from financial markets to AI development was not accidental. Divyaprakash Jha spent ten years in the Gold and FX markets, a period that shaped his approach to automation and risk management.

During this decade, Jha built over 500 automated trading systems, known as Expert Advisors (EAs), for the MetaTrader 5 platform. These systems were essentially "small agents": they read market signals, made decisions, and executed trades without human intervention. They had to be disciplined, they had to follow strict rules, and most importantly, they had to be verifiable to avoid losing money.

This experience provided the perfect blueprint for modern AI. The core questions Jha asked in trading are the same questions he asks when building AI agents today:

  • What does the system know?
  • What is it allowed to do?
  • How do you check its work before it causes a loss?

The evolution from writing rules for trading to building agentic AI workflows is a natural progression of automating complex decision-making processes. You can read more about this journey in this story of the founder's journey.

Questions people ask

Where is the AI startup located?

The studio, led by Divyaprakash Jha, is based in Palghar, Maharashtra, India.

What kind of AI products are being built in Palghar?

The studio builds decision intelligence engines (GodEngine), AI assistants for Indian business workflows (AskShiva), and agentic workflow platforms (Shiva Agents).

How many people work at the studio?

The studio operates with a team of three human operators and a fleet of fifteen AI agents.

Does the AI startup use paid APIs?

The studio runs its own GPU and CPU servers with open models so that bulk AI work does not depend on paid APIs.

What is 'agentic AI' in the context of these products?

It refers to AI agents that can plan, use tools, verify their own work, and execute multi-step tasks like creating GST invoices or managing workflows.